ChorusTIC: Training-Free Multivariate Time Series Classification via Chorus In-Context Learning
Abstract
Time series classification underpins applications in healthcare, sensing, and industrial monitoring. Although time series foundation models support forecasting and transferable representation learning, classification still typically requires fitting a task-specific classifier on each target dataset, while individual channels of multivariate inputs are often encoded independently. We introduce ChorusTIC, a classification-native foundation model for in-context classification across heterogeneous channel configurations without target-task parameter updates. ChorusTIC combines episode-consistent Random Subchannel Slot Concatenation with a shared dual-axis encoder to model temporal and cross-channel interactions and map variable channel configurations into a fixed-width representation independent of the original channel count. It then calibrates feature axes using context-derived distributions and predicts query labels through leakage-protected in-context learning. We pretrain ChorusTIC solely on synthetic labeled episodes comprising context and query sets that share a task background, with classes distinguished by sparse temporal or cross-channel rules. Evaluations on the complete UEA-30 and UCR-128 archives show strong full-context and low-label performance without target-specific classifier fitting.
Related
- Class-Conditioned Diffusion Augmentation and Explainable In-Context Ensemble Learning for Imbalanced Maintenance Action Classification in Aircraft Systems
- Comparing the use of supervised machine learning variable selection methods in the context of two-group classification in the psychological and health sciences
- Bayesian Matrix-Valued Graphs for Context-Dependent Multivariate Relationships
- Multivariate analysis of poverty and social exclusion across EU countries in the context of Sustainable Development Goal 1
- In-context classification: Calibration and noise in linear attention with softmax readout
BibTeX
@misc{fang2026chorustic,
title = {{ChorusTIC: Training-Free Multivariate Time Series Classification via Chorus In-Context Learning}},
author = {Juntao Fang and Shifeng Xie and Ruichu Cai and Shengji Zheng and Zijian Li and Keli Zhang and Lujia Pan and Themis Palpanas and Zhifeng Hao},
howpublished = {arXiv},
year = {2026},
month = sep,
doi = {10.48550/arxiv.2608.24033},
eprint = {2608.24033},
archivePrefix = {arXiv},
url = {https://arxiv.org/abs/2608.24033},
}
RIS
TY - UNPB
TI - ChorusTIC: Training-Free Multivariate Time Series Classification via Chorus In-Context Learning
AU - Juntao Fang
AU - Shifeng Xie
AU - Ruichu Cai
AU - Shengji Zheng
AU - Zijian Li
AU - Keli Zhang
AU - Lujia Pan
AU - Themis Palpanas
AU - Zhifeng Hao
PY - 2026
DA - 2026/09/30
PB - arXiv
DO - 10.48550/arxiv.2608.24033
UR - https://arxiv.org/abs/2608.24033
AB - Time series classification underpins applications in healthcare, sensing, and industrial monitoring. Although time series foundation models support forecasting and transferable representation learning, classification still typically requires fitting a task-specific classifier on each target dataset, while individual channels of multivariate inputs are often encoded independently. We introduce ChorusTIC, a classification-native foundation model for in-context classification across heterogeneous channel configurations without target-task parameter updates. ChorusTIC combines episode-consistent Random Subchannel Slot Concatenation with a shared dual-axis encoder to model temporal and cross-channel interactions and map variable channel configurations into a fixed-width representation independent of the original channel count. It then calibrates feature axes using context-derived distributions and predicts query labels through leakage-protected in-context learning. We pretrain ChorusTIC solely on synthetic labeled episodes comprising context and query sets that share a task background, with classes distinguished by sparse temporal or cross-channel rules. Evaluations on the complete UEA-30 and UCR-128 archives show strong full-context and low-label performance without target-specific classifier fitting.
ER -